Crop-yielding Prediction using Neural Network for Stochastic Differential Equation Parameters Estimation
The project will predict the crop yielding by defining new SDE parameters that can accurately be estimated in SDE model using a neural network under specific noise level regimes to forecast the crop yielding.
Project status
Content navigation
About
General Overview: -
The project will predict the crop yielding by defining new SDE parameters that can accurately be estimated in SDE model using a neural network under specific noise level regimes to forecast the crop yielding.
Domain Overview: -
This project employs a collection of stochastic differential equations (SDE) in Itô form using an artificial neural network (ANN) to predict the parameters of the stochastic differential equation of the chaotic time series of wheat yielding daily, monthly, and yearly.
The SDE model will have new functions for the drift and diffusion SDE parts. The Euler and Stratonovich methods evaluate the output model. The Neural Network will be used to estimate the parameters to reach the best output using a gradient.
Technical Overview: -
Given one-dimension time-homogeneous SDE
dX=μX;θdt+g(X;θ)dW (1)
the task is to estimate the parameter θ from a sample of (N+1) observations X0,….., Xn of the process at known times t0,….., tn. In equation (1), dW is the differential of the Wiener process (Brownian motion). μX;θ is the instantaneous drift , g(X;θ) is the instantaneous diffusion
The SDE model will propose new functions for the drift and diffusion SDE parts. The output model is evaluated by Euler and Stratonovich methods.